Lithographic Additive Manufacturing of metal-based suspensions
A paper from Gerald Mitteramskogler, Martin Schwentenwein and Simon Seisenbacher (Lithoz GmbH, Austria), Carlo Burkhardt (Pforzheim University of Applied Sciences, Germany), Oxana Weber (OBE GmbH & Co. KG, Germany) and Christian Gierl-Mayer (Technical University of Vienna, Austria) presented a study on the lithographic AM of metal-based suspensions . Stereolithography was the first commercialised AM technology, for the building of polymeric prototypes by a spatially controlled solidification of a liquid resin by photo-polymerisation, and was invented by Chuck Hull in the mid-1980s. Following the definitions in ISO/ASTM 52900, similar processes are now categorised under ‘Vat Polymerisation’.
Industrial-scale vat polymerisation processes are already applied for the shaping of specialised materials, such as toughness-modified plastics or high performance ceramics. As the AM process is a layered process, a certain level of translucency of the suspension is required to attach the layers on top of each other and to ensure sufficient mechanical strength of the manufactured part.
Within particle-filled photo-reactive suspensions, the penetration of light is highly governed by the degree of absorption of the powder particles (a material constant), the filler content of the suspension (usually around 50 vol.%), the physical size and shape of the particles and the scattering of the light at the particle/binder interface (i.e. the difference in refractive indices).
To-date, these limitations have hindered the exploitation of the potential benefits of the vat polymerisation AM technology for the shaping of high quality, metal-based green parts, because metal powder is optically dense and shows a high level of absorption.
This paper introduced a novel vat polymerisation process, developed by Lithoz GmbH during a European Commission Horizon 2020 project, REProMag, which is suitable for the manufacture of highly complex green parts based on 316L stainless steel. The issues previously highlighted have been solved by developing a novel slurry composition and a new generation of AM machines. The aligned interaction of machine and material allows the application of thin layers of material with precision down to the micrometre level.
A photo-reactive suspension was prepared based on commercially available di- and polyfunctional methacrylates (60 wt.%). The slurry included an initiation system with a proprietary photo-initiator, which absorbs light in the specific wavelengths emitted by the light engine. Using this formulation, a solid loading of 316L powder up to 50 vol.% was achieved. A homogeneous mixture was prepared by centrifugal mixing, with both the organic and the metallic powder being added in the cup and homogeneously dispersed.
A novel AM machine was developed based on the principle of vat polymerisation (Fig. 2). The liquid starting material was polymerised from above by a high-performance projection unit. The building platform with the green parts was lowered layer-by-layer, according to the set layer thickness. For the 316L material, a thickness of 50 μm was chosen. After the curing of a layer, the wiper blade applies a fresh film of material. The size of the building platform was 75 x 43 mm and the resolution in the X and Y directions was 40 μm. The printing time of a single layer was 35 s, which resulted in a build speed of 6 mm/h in the Z direction (or around 20 cm³/h in volume terms).
The self-supporting function of the material enabled volume-optimised placement of different parts on a single building platform without the need for additional support structures. The only manual operation necessary was to load the parts into the respective program (Netfabb by Autodesk® or Magics by Materialise). The 3D nesting operation was automatically performed by the software. The print job in Fig. 3 was generated, resulting in volume occupancy of 18% of the overall building volume. The minimum distance between the parts was set at 1 mm. However, smaller values down to 0.1 mm are possible. The software slices the parts to create the image information for each layer.
Thermal debinding of the three-dimensional green parts was performed in an inert atmosphere up to 320°C to avoid oxidation. Fig. 4 shows the temperature program used, which was optimised for a cylinder of 10 mm in diameter and height. The debinding was optimised by thermo-mechanical and thermo-gravimetric analyses. After the debinding, the parts were sintered in a hydrogen atmosphere at Ohnmacht & Baumgärtner GmbH & Co. KG (OBE). The maximum sintering temperature was 1360°C with a 2 h holding time.
To evaluate the mechanical strength of the sintered parts, tensile test specimens (Fig. 5) were manufactured and sintered. The geometry of the parts was based on DIN EN ISO 6892-1, with the test length being 36 mm and the test area 6.5 x 3 mm. During manufacturing, the parts were oriented parallel to the building platform.
The tensile tests were performed under a quasi-static load using a universal testing machine at ambient conditions (20°C, laboratory air). It was shown that a yield strength of 233 ± 26 MPa and a tensile strength 502 ± 13 MPa could be achieved. The density of the samples was around 96% of the theoretical density of 316L. The additively manufactured material performed similarly to traditionally processed 316L, which showed a yield strength 220 MPa and a tensile strength 530-680 MPa.
Tensile data are only available for a single build direction, so, to evaluate any dependencies on the build direction of the green part, further studies need to be conducted. Also, the fatigue strength of the material will be a topic for further studies.
Fig. 6 shows an etched micrograph of a sintered 316L sample. The remaining porosity is closed, which would allow further densification by Hot Isostatic Pressing. This micrograph also confirms the measured density levels.
Fig. 7 shows sintered parts printed with the developed material and process. After the AM process, the parts have been cleaned and the excess material could be recycled for reuse. Due to the densification process, debinding and sintering caused a homogeneous shrinkage of about 22%. Slight distortions due to gravitational forces during high temperature sintering are visible. To minimise these distortions, the structures could be placed on a customised setter plate, such as a machined porous alumina material.
These early results demonstrate the great potential of the LMM process to be used either as a complementary technology to MIM, or other AM processes such as Powder Bed Fusion or Binder Jetting. It shows high advantage when focusing on smaller, more complex parts that require a high feature resolution and improved surface quality. The process can be used to directly produce parts in a small-scale series or to manufacture prototypes prior to a MIM-based mass series production. Since the final geometry of the part is developed by means of a sintering process, similar microstructural characteristics compared with MIM can be achieved.
Characteristics of porous Ti6Al4V materials produced by three-dimensional fibre deposition and bio gel casting
In the second paper, Marleen Rombouts, Steven Mullens and P Weltens (Vito, Belgium) compared the processes of three dimensional fibre deposition (robocasting) and bio gel casting for the manufacture of porous Ti6Al4V structures.
For biomedical implant applications, the porosity level of Ti6Al4V is varied to control the material’s elastic modulus (to match as closely as possible that of bone to avoid stress shielding) and to provide interconnected porosity for the in-growth of new bone tissue. The optimum pore and interconnection for bone in-growth has been reported as being in the range 100-500 µm.
Robocasting is an AM technology originally developed at Sandia National Laboratories in the 1990s and is particularly suited to the production of structured porous materials. Some characteristics of robocasting technology, compared to other more common metal AM technologies such as Selective Laser Melting, are:
- It is an indirect process, in which, after printing, the parts are densified by a thermal process
- There is no powder bed and thus cleaning procedures for the removal of powder from internal cavities are avoided
- Good mechanical properties can be achieved, with an absence of residual stresses, as the shaping is performed at room temperature and specific furnaces are used for calcination and sintering in controlled atmosphere
- The possibility is offered to induce surface roughness and micro-porosity within the fibre by a phase inversion technique
- The technique is applicable to all powder materials
- The technology is somewhat more limited in porous design compared to powder bed based AM technologies.
Gel casting is a direct foaming technique which starts from a water-based slurry of titanium powder to which a foam-forming, gelling and stabilising agent and a binder are added.
Spherical Ti6Al4V powder, produced by Electrode Induction-melting Gas Atomisation and finer than 45 μm (d50 ~ 30 μm), was used in both the robocasting / 3D Fibre Deposition process (3DFD) and gelcasting process studies. 3DFD is an AM method based on the concept of micro-extrusion combined with computer controlled movement in three dimensions. The core in this process is the highly loaded paste, which is extruded through a nozzle. The typical flow chart for the 3DFD technology is shown in Fig. 8. Parts are printed with a nozzle with a diameter of 0.4 mm and layer thickness of 0.3 mm. The last step in the flow chart comprises a thermal treatment, in which the parts are calcined to remove the binder from the paste and subsequently sintered in an argon atmosphere. The samples are first heated at 0.5°C/min to 500°C and, after a dwell time of 1 h, are heated to 1350°C with a holding time of 1 h.
The bio gel casting process is shown schematically in Fig. 9. Different porosities and pore size distributions have been obtained by changing the mechanical mixing time. The same heating cycle for debinding and sintering, as for the 3DFD samples, has been applied.
The products resulting from the two processes have been compared through structural characterisation and static compressive mechanical property measurements.
In terms of structural comparisons, gel cast Ti6Al4V foams are characterised by a broad pore size distribution (Fig. 10) and an average pore size in the range of 50 – 200 µm, whereas robocast metals are characterised by a well-controlled porosity and a narrow pore distribution (Fig. 11).
The overall average width of robocast fibres, produced with a nozzle diameter of 0.4 mm, is 0.35 mm. The standard deviation of the fibre diameter is 0.02 mm (i.e. around 5%). By changing the fibre spacing between 0.8 mm and 1.2 mm, the final relative density varies from 54.5% to 27.6%. The volumetric shrinkage during sintering of the robocast structures is 33.9 ± 0.6%.
The compressive mechanical behaviour of gel cast materials is illustrated in Fig. 12. The foams show a higher strength and stiffness at a higher density. Stiffness values in the range 2-4 GPa are obtained (cf. trabecular bone: 0.2-2 GPa). Compressive strengths in the range 45-100 MPa are obtained (trabecular bone is in the range of 15-50 MPa) at relative densities of 20-35%.
For robocast material, a higher spacing between fibres results in a lower stiffness and compressive strength (Fig. 13). The stacking design has a smaller effect on the mechanical behaviour than the scan spacing. The stiffness and strength upon loading along the layers are significantly larger than those measured perpendicular to the layers, especially at higher scan spacing. Fibres stacked twice on top of each other (design 113355) lead to a higher stiffness and strength.
Additive Manufacturing of metal parts using Fused Filament Fabrication
Finally, a poster, associated with this session and presented by a Fraunhofer team of Sebastian Riecker, Sebastian Boris Hein, Thomas Studnitzky, Olaf Andersen and Bernd Kieback, focused on the AM of metal parts by Fused Filament Fabrication (FFF). FFF is one of the most popular and most widely used AM technologies because of its relative simplicity and, therefore, low investment cost for the machines.
During the printing process, a filament with a diameter of 1.75 mm to 3 mm is melted and extruded through a small nozzle of typically 0.4 mm to 0.8 mm in diameter. Controlling the nozzle’s movement, the extruded strand is printed in a defined geometry to form the component layer by layer. The filament, being the printing material, also acts as the push rod for the extrusion. It is moved by a driving wheel and has to transfer the driving force to the liquified filament material in the nozzle (Fig. 14). Therefore, the filament must have sufficient strength to be processed in the printing machine. While being flexible enough to be coiled, strength and stiffness is necessary to act as a push rod for the filament extrusion without buckling.
In general, a low viscosity at the process temperature is necessary for the extrusion through the small nozzle and a high filament stiffness is favourable. In addition, the printed strands must have good adhesion to the printing substrate and the already printed structures and show little warpage due to thermal expansion.
To allow debinding and sintering after printing, either two different polymer materials with different decomposition temperatures must be used in the feedstock, or an otherwise removable phase (e.g. wax) must be present in addition to the polymer.
Filled thermoplastic filaments that feature improved physical, optical or haptic properties by using reinforcements such as carbon fibres, wood fibres or metal or ceramic powders have been developed. However, the volume content of the filling materials lies mainly between 10 and 40 vol.%, making them unusable for further sintering processing steps. Only a few research studies have, to-date, focused on filament materials that can be debound and sintered following the printing process to achieve a dense ceramic or metal part.
The presented work was, therefore, aimed at developing a feedstock material that can be printed, debound and sintered to full density. The studies focused on three different material types for the filament production: PLA, PA and Polypropylene PP. Dispersing agents and softeners were used to modify the rheology of the melt as well as the physical properties and debinding performance. 316L stainless steel powder with a D50 of 7 μm was the metal powder used as filling material throughout this work. An overview of the tested filaments is provided in Table 1 by showing examples of the filaments in each experiment set, although the exact feedstock composition was kept confidential.
The filaments were produced by two different manufacturing routes. In the first route, a twin screw extruder with six heating zones was used to compound and extrude the feedstock mixtures to filaments with a diameter of 3 mm. The kneading temperature given in Table 1 refers to the heating zones 1-5 of the extruder, the extrusion temperature refers to heating zone 6. In the second route, a kneader and a table extruder were used to produce filaments with a diameter of 1.75 mm.
The filaments showed a large variance in quality between the different polymer types used in the study, an overview being given in Table 2. As anticipated, the physical properties of the polymer matrix had a large influence on the filament properties. Brittle materials tended to become significantly weak at high volume loadings (F1), whereas the properties were maintained up to volume loadings as high as 59 vol.% for flexible and ductile feedstock materials (F4, F5). According to these results, the PLA samples were too brittle and the PP samples were too flexible to perform printing trials. The addition of a softener had only a small impact on the tested flexible PP filaments and led to a significant decrease of strength of the PA filaments. Surface quality of the filaments and the extruded strand was equally influenced by the material composition and the process parameters. Fig. 15 shows the surface of filaments F4 and F3. A similar difference in surface quality could be observed on filament F4 with different extrusion temperatures. Generally, the filament extrusion gave the first insights into feedstock behaviour. Artefacts, such as bearding, shape loss or low surface quality, were reproduced with a small printing nozzle. In addition to the observations of the extruded filament, the extrusion torque and flow rate of the extruded filament were used as an orientation value for rating the extrusion behaviour. Materials with a lower value of extrusion torque showed better extrusion behaviour through the small printing nozzle.
A lower temperature and a medium extrusion speed led to the best results in the tests. Examples of the results can be found in Fig. 16 for filament F2 using a nozzle size of 0.8 mm. A good surface quality of the filament and of the extruded strand was found to be important for the avoidance of large pores in the printed parts.
Printing trials were performed on filament type F2 using a commercial printer that used a Bowden cable for feeding of the filament. A temperature of 235°C with a printing speed of 25 mm/s was set for the trials. Sintering experiments were performed on chemically and thermally debound filaments as well as printed parts at temperatures of 1300°C to 1350°C under a hydrogen atmosphere.
With the selected printer, only filaments of type F2 could be printed due to the high filament strength that is required for the Bowden cable feeding system. Due to the low melt strength, the retraction setting was set to very low values and a layer height of 0.1 mm was set in the example given in Fig. 17. The box could be printed with a wall thickness of about 1.5 mm and a side length of 1 cm. After thermal debinding and sintering, the walls show a small warpage caused by the shrinkage of about 18% and the box’s geometry without a dense bottom layer. As can be seen in the cross section, the remaining porosity was still high at around 10% in this example. However, no layering effect could be found within the dense compact in the investigated cross sections. It can therefore be assumed that the anisotropy in mechanical strength found for plastic components can be minimised in printed and sintered parts. Furthermore, filaments of type F4 could be sintered to almost full density of 99% under a hydrogen atmosphere with a partial pressure of 850 mbar and a sintering temperature of 1350°C. A thermal debinding route was performed under the same atmospheric conditions between room temperature and 500°C, leading to a slight deformation of the filament. Printability still has to be validated. Due to the flexibility of the filaments, a printer setup with the driving wheels above the nozzle is required in this case.
Work undertaken between the conference manuscript submission and the preparation of the conference poster recorded significant progress in the refinement of the technology. The latest filament composition has a content of 60 vol.% of metal powder and shows good printability. The green parts can be debound in ethanol, isopropanol or acetone before thermal debinding and sintering to relative densities up to 98%, with a linear sintering of shrinkage of around 16% without any major printing defects despite the layer-by-layer printing process (Fig. 18).
The printability and sinterability of the filament was demonstrated on more complex parts, such as the twisted box shown in Fig. 19, with good dimensional stability after sintering. The authors stated that the next steps include the design and construction of an integrated process line for the printing, debinding, sintering and CNC machining of complex metal parts.
The authors summarised their findings by concluding that the FFF process with subsequent debinding and sintering can be realised with highly filled filaments. Low investment costs and the high complexity of printable designs makes this manufacturing method interesting both for small companies and small batch sizes.
 Lithographic Additive Manufacturing of Metal-based Suspensions, G Mitteramskogler, M Schwentenwein and S Seisenbacher, C Burkhardt, O Weber and C Gierl-Mayer, as presented at Euro PM2017 Congress and Exhibition, Milan, Italy, October 1-5, 2017, and published in the Conference Proceedings by the European Powder Metallurgy Association (EPMA).
 Characteristics of Porous Ti6Al4V Materials Produced by Three-Dimensional Fiber Deposition and Bio Gelcasting, M Rombouts, S Mullens and P Weltens, as presented at Euro PM2017 Congress and Exhibition, Milan, Italy, October 1-5, 2017, and published in the Conference Proceedings by the European Powder Metallurgy Association (EPMA).
 3D Printing of Metal Parts by Means of Fused Filament Fabrication – A Non Beam-Based Approach, S Riecker, S B Hein, T Studnitzky, O Andersen and B Kieback, as presented at Euro PM2017 Congress and Exhibition, Milan, Italy, October 1-5, 2017, and published in the Conference Proceedings by the European Powder Metallurgy Association (EPMA).
Dr David Whittaker
34 Dewsbury Drive
Tel: +44 1902 338498
Email: [email protected]
Euro PM2017 Proceedings
The full proceedings of the Euro PM2017 Congress is now available to purchase from the European Powder Metallurgy Association. Topics covered include:
- Additive Manufacturing
- PM Structural Parts
- Hard Materials & Diamond Tools
- Hot Isostatic Pressing
- New Materials & Applications
- Powder Injection Moulding